NTNU Norwegian University of Science and Technology Faculty of Information Technology and Electrical Engineering Department of Electric Power Engineering
Anders Ulfsnes
Investigating the Effect of Eccentric Conductor Positions on a Rogowski Coil in Laboratory and Finite Element Method Simulations
Master’s thesis in Energy and Environmental Engineering Supervisor: Irina Oleinikova
Co-supervisor: Mohammad Khalili Katoulaei, Hans Kristian Høidalen
June 2021
Master ’s thesis
Anders Ulfsnes
Investigating the Effect of Eccentric
Conductor Positions on a Rogowski Coil in Laboratory and Finite Element
Method Simulations
Master’s thesis in Energy and Environmental Engineering Supervisor: Irina Oleinikova
Co-supervisor: Mohammad Khalili Katoulaei, Hans Kristian Høidalen June 2021
Norwegian University of Science and Technology
Faculty of Information Technology and Electrical Engineering
Department of Electric Power Engineering
Sammendrag
Endringer i strukturen til kraftsystemet gir nye muligheter og utfordringer for drift og vern av nettet. De siste ˚arene har det kommet digitale nettstasjoner, men disse installasjonene skjer ikke over natten, og det kreves derfor nye flerteknologiløsninger. Rogowskispolen er en lovende strømsensor som har fordeler i verninstallasjoner. En stor fordel er at luftk- jernen aldri mettes i motsetning til konvensjonelle strømtransformatorer med jernkjerne.
Samtidig er det utfordringer som m˚a løses. Den p˚avirkes av flere faktorer og i løsninger med differensialbeskyttelsesrel´eer kan dette for˚arsake uønskede feil.
M˚alenøyaktigheten til en rogowskispole testes i et laboratorieoppsett. Posisjonen p˚a led- eren endres for ˚a undersøke hvordan dette p˚avirker utputten. Resultatene viser ingen
˚apenbare trender i utputten n˚ar lederposisjonen endres, men generelt endres feilen n˚ar lederen flyttes bort fra sentrum. Feilen til rogowskispolen som testes er generelt ganske høy med hele1,6%feil i sentrert posisjon. Feilen endres deretter med opptil0,25pros- entpoeng n˚ar lederen flyttes ut fra sentrum.
En modell lages med endelige elementers metode for en rogowskispole og konfigureres til ˚a etterligne den som ble brukt i eksperimentet. Modellen lages i 3D i en flerfysikksplat- tform. Modellen testes i ulike konfigurasjoner og parametere. Først har spolen16viklinger, og feilen som m˚ales i forhold til den sentrerte lederposisjonen er mye høyere enn de som ble m˚alt i laboratoriet. Feilene viser ogs˚a trender i hvordan lederposisjonen p˚avirker feilen.
For ˚a forsøke ˚a gjenskape resultatene fra eksperimentet blir det forsøkt ˚a gjøre viklingstet- theten mer tilfeldig, slik at den i større grad kan gjenspeile hvordan viklingene blir laget i virkeligheten. Resultatene blir noe endret, men etterligner ikke resultatene fra eksperi- mentene i særlig grad. Antall viklinger blir s˚a økt til48og testene repeteres. De resul- terende feilene er n˚a cirka like store som de som ble m˚alt i laboratoriet.
Laboratorieoppsettet har flere begrensninger som bør fikses. Deretter kan forbedrede tester brukes til ˚a verifisere modellen. Bruken av en endelige elementers metode-modell er en verdifull metode for ˚a forske p˚a rogowskispolen ettersom man enkelt kan undersøke ikke- ideelle konfigurasjoner. Modellen kan ogs˚a utvides med en integrator og kompensering- steknikker. Dette gir flere muligheter for ˚a utvikle løsninger som gjør at rogowskispolen en brukbar sensor i flerteknologiløsninger i vern og for p˚alitelig operasjon av systembeskyt- telsesløsninger.
Abstract
Changes in the power system structure create new possibilities and challenges in the pro- tection and operation of the grid. Digital substations are emerging, but these installations are not done overnight, and hence the transitions require new thinking in terms of inter- operability in protection applications. The Rogowski coil (RC) is a current sensor which offers benefits in protection schemes. Preeminently that the air core does not saturate, compared to conventional iron core Current transformers (CTs). But challenges limit its current applications. It is affected by several influencing quantities and in differential pro- tection schemes this can cause malfunctioning.
A test rig is developed, and the measurement accuracy of a RC is tested in the setup.
The conductor position is adjusted to investigate the effect on the output. The overall accuracy is about1.6%error in the centered position. The accuracy changes up to0.25 percentage points with the eccentric configuration. The RC output varies based on where the conductor is located, but trends in how conductor adjustments affect the error is not obvious.
A Finite element method (FEM) model is created for a RC configuration similar to the one used in laboratory tests. The model is created using 3D geometry in a multiphysics environment. The model is tested with several configurations and parameter changes. Ini- tial tests with16turns result in composite errors with a magnitude which is a lot higher than what is achieved in the laboratory tests. To investigate how randomness in the wind- ing distribution affects the output, an inhomogeneous winding distribution is used. The resulting output is only slightly changed compared to the homogeneous test. With48ho- mogeneously distributed turns the error resembles the values achieved in the lab, but with clear patterns related to how the conductor adjustments affects the error.
The laboratory setup has several limitations which should be improved. It can then be used to verify the FEM model. Using a FEM model is a valuable method of researching the RC as it can assess nonideal configurations. It can also be extended to utilize an inte- grator and compensation techniques which can be designed to improve the accuracy. This can make the RC a viable current transducer in interoperable protection applications and reliable System protection scheme (SPS) operation.
Preface
The author wants to thank PhD candidate Mohammad Khalili Katoulaei who helped a lot with the organization of the laboratory tests as well as inputs and help with the simulation model. Mr Katoulaei has been very helpful and available at most hours during the week.
Thanks to supervisor professor Irina Oleinikova for guidance and follow ups during the process of the master thesis, for inputs on the structure and contents in the thesis and for feedback on the work conducted. Thanks to Aksel Andreas Reitan Hanssen for gener- ous support and logistics help related to obtaining devices for use in the laboratory work.
Thanks to Anders Gytri for IT support related to server and desktop simulations.
The work leading up to the thesis has spanned one year. The first part of the work focused on preliminary literature studies, and basic modeling of RCs using lumped parameters.
The second part goes further into the performance of the RC. Materials from the first part of the work is reused in the thesis as it gives a background in the uses of the RC in pro- tection applications ans system protection schemes. These parts are: chapter 1 including Trends in the Power System, chapter 2 includingSystem Protection SchemesandDiffer- ential Protection, as well as the introduction to chapter 3 and its first sectionModeling &
Output Integration.
Table of Contents
List of Tables i
List of Figures vii
List of Abbreviations x
1 Introduction 1
1.1 Trends in the Power System . . . 2 1.2 Scope of Thesis . . . 4
2 Power System Protection 7
2.1 System Protection Schemes . . . 11 2.2 Differential Protection . . . 14 2.3 Prodig Perspective on Digital Substations . . . 17
3 Rogowski Coil Modeling & Testing 19
3.1 Modeling & Output Integration . . . 20
3.2 Rogowski Coil Dependence on Influence Quantities . . . 23
3.3 Laboratory Setup . . . 27
3.4 Finite Element Method Modeling Setup . . . 29
4 Laboratory Testing & Finite Element Method Simulations 33 4.1 Laboratory Tests . . . 34
4.1.1 Horizontal Adjustments of Conductor Position . . . 36
4.1.2 Vertical Adjustments of Conductor Position . . . 36
4.1.3 Diagonal Adjustments of conductor Position . . . 37
4.1.4 Influence from External Conductors . . . 39
4.2 Finite Element Method Modeling Simulations . . . 39
4.2.1 Conductor Positions for16Turns Configuration . . . 42
4.2.2 Conductor Positions with Inhomogeneous Winding Distribution . 44 4.2.3 Conductor Positions for48Turns Configuration . . . 45
4.2.4 Effect of Discontinuity . . . 47
4.2.5 Effect of Output Resistance . . . 48
4.3 Summary and Discussions . . . 49
4.3.1 Laboratory Tests . . . 49
4.3.2 Finite Element Method Modeling . . . 51
4.3.3 Common Discussions . . . 52
5 Conclusions & Further Work 55 5.1 Further Work . . . 56
References 57
Appendix A Laboratory Tests Supplements 67
Appendix B Finite Element Method Simulations Supplements 75
List of Tables
3.1 Physics nodes used in themagnetic fieldsinterface in the model . . . 31 3.2 Mesh statistics for the model when the conductor is in its centered position 31
4.1 Oscilloscope measured RC and CT RMS values from vertical adjustments 34 4.2 Oscilloscope measured RC and CT RMS values from horizontal adjustments 35 4.3 Oscilloscope measured RC and CT RMS values from diagonal adjust-
ments. H. pos - horizontal position and V. pos - vertical position . . . 36 4.4 Materials used in RC model . . . 40 4.5 Output voltage RMS ratio to case with δ = 1° and composite error for
x, y= 0,20mm conductor position for different discontinuities. . . 50 4.6 Output resistance and measured voltage and current in RC. . . 50
B.1 X- and y-coordinates for different conductor positions in a parametric sweep simulation. . . 77
List of Figures
1.1 The traditional grid compared to IEEE’s vision of the smart grid structure 3
2.1 An overview of an IEC 61850 based substation . . . 8
2.2 Normal condition and heartbeat messages for GOOSE packet transmission 10 2.3 A system protection terminal . . . 12
2.4 A novel protection architecture in smart grids . . . 13
2.5 Differential relay connection diagram . . . 15
2.6 Characteristics of a level detector . . . 16
2.7 Overlapping protection zones . . . 16
2.8 Schematic of technology used and intended interconnection of communi- cation withing ProDig project laboratory resources . . . 18
3.2 A rogowski coil with an R-C integrator . . . 22
3.3 Bode diagrams of two different integrators with the same value of R and C 23 3.4 The discontinuity angle, β, to describe the angle difference between the first and last turn in the RC . . . 24
3.5 Position of angled conductor in laboratory tests . . . 25
3.6 Temperature dependence in RC with different load resistances . . . 26
3.7 The Fluke i2000 flex used in laboratory testing . . . 27
3.8 Photos from the laboratory setup and the equipment used . . . 28
3.9 The RC model in orange with the air domain around the coil and the con- ductor in the middle of the figure. The terminals of the RC is seen on the right. . . 29
3.10 Magnetic field in a simple coil model showcasing the effect of infinite el- ement domains. Infinite element domains are not added to the left and right side due to terminal connections of the conductor and the coil. Cur- rent is excited in the coil on the right side, and induces a voltage in the U shaped conductor. The blue color shows the magnetic field strength, and the curves shows the direction of the magnetic field . . . 32
4.1 The RC in the test rig frame with side references indicated with text. The RC opening is the black part on the right side. The conductor is seen as the grey bar in the middle . . . 33
4.2 Laboratory test percentage composite error of RC for horizontal adjust- ments of conductor position. . . 34
4.3 Laboratory test percentage composite error of RC for vertical adjustment of conductor position . . . 35
4.4 Laboratory test percentage composite error for diagonal adjustments from bottom left to top right of RC window . . . 37
4.5 Laboratory test percentage composite error for diagonal adjustments from bottom right to top left of RC window . . . 37
4.6 Heatmap showing the percentage ratio between the measured RMS of the RC and the CT for different conductor positions. A high value above 100 indicates that the RMS for the current conductor position is higher than the CT RMS current . . . 38
4.7 Current output from RC with no current in any conductors, and with30A in external conductor adjoined with RC insulation . . . 39
4.8 Magnetic field distribution in COMSOL RC model. Model shown from XY-plane. Conductor in centered position with1° discontinuity . . . 40 4.9 Test numbers of conductor positions in RC window in the simulation model 41 4.10 Composite error for all 29 conductor positions in subfigures showing a
sweep from one side to the other with the centered position in all figures.
Homogeneous winding distributed with1° discontinuity . . . 42 4.11 Ratio between the measured RMS voltage of a position compared to the
centered position in per cent . . . 43 4.12 16turn inhomogeneous winding distribution seen from xy-plane for RC
withδ= 4°. No return wire . . . 44 4.13 Composite error for all 29 conductor positions in subfigures showing a
sweep from one side to the other with the centered position in all figures.
16turn inhomogeneous winding distributed with4° discontinuity. No re- turn wire . . . 45 4.14 The geometry of the RC in orange with 48 turns and return wire. Grey
conductor in the centered position. Discontinuity shown in the top right corner with the output terminals . . . 46 4.15 Composite error for all 29 conductor position adjustments for a RC with
d = 15resulting in48turns. Homogeneous winding distributed with1°
discontinuity . . . 47 4.16 The RC geometry with different discontinuities seen from XY-plane. The
discontinuity is located on the right side . . . 48 4.17 Measured RC voltage for different output resistances for a sample period . 51
A.1 Schematic of the RC accuracy test setup. The schematic shows the con- nections, devices and the cable types in the setup. . . 67 A.2 Horizontal adjustment in the test rig. The ruler is used to show the place-
ment of the conductor and is indicated by the arrow. The black vertical cylinder in the background is the conductor stand. Adjustments are done by unscrewing the bolt in the top right of the photo. . . 68
A.3 Connection of the terminals to the Omicron current output using banana plugs. The current is outputted through current output A terminal one and returned on the neutral terminal. The schematic in figure A.1 uses the same color scheme, but the external conductor (bottom) is not connected. 69 A.4 The banana plug connected to the aluminium conductor with the RC in the
background. . . 69 A.5 Closeup of the RC discontinuity. The locking mechanism is located just
above the strip. . . 70 A.6 The vertical adjustment of the conductor position. Unscrewing the screw
enables vertical adjustment of the frame. Since the frame is held up by the cylindrical arm, it means the angle of the frame is also adjustable, which makes it difficult to make proper vertical adjustments with no other parameter changes. . . 70 A.7 Protractors for adjustment of conductor angle. Arrow indicator indicates
the angle. . . 71 A.8 Adjustment of angle of lower cylindrical arm for adjustment of both con-
ductor position and angle simultaneously. . . 71 A.9 Reference Pearson Electronics inc. current monitor used for comparisons
with the RC. It measures the output of the Omicron. . . 72 A.10 Cables from ITECH connected to conductors on test rig. . . 73 A.11 Per unit current output of ITECH over time for different output levels. . . 74
B.1 Direction of numerical calculated wire in conductor (blue) and RC (red) indicated by arrowheads after the initial coil geometry analysis. If the initial coil geometry analysis fails, the current directions does not follow the coil windings. . . 76 B.2 A comparison of the vertical adjustments of the conductor position in the
laboratory tests, and the simulations with an inhomogeneous winding dis- tribution. . . 78 B.3 Phase shift from different conductor positions when adjusted horizontally.
Homogeneous winding distribution with48turns. . . 78
B.4 Phase shift from different conductor positions when adjusted vertically.
Homogeneous winding distribution with48turns. . . 79 B.5 Phase shift from different conductor positions when adjusted diagonally.
Homogeneous winding distribution with48turns. . . 79 B.6 Phase shift from different conductor positions when adjusted diagonally.
Homogeneous winding distribution with48turns. . . 80 B.7 Current density in the coil indicated by red arrows and in the conductor
indicated by blue arrows. current direction is directed into the return wire, and then through the windings.1Ωoutput resistance andδ= 1°. . . 81
List of Abbreviations
ANN Artificial neural network CB Circuit breaker
CT Current transformer DSP Digital signal processing
FACTS Flexible alternating current transmission system FEM Finite element method
GOOSE Generic object oriented substation events GPS Global positioning system
HVDC High voltage direct current IED Infinite element domain IoT Internet of Things MAC Media access control MU Merging unit
NCIT Nonconventional instrument transformer OpAmp Operational amplifier
ppm Parts per million PT Potential transformer
PV Photovoltaic RC Rogowski coil
SPS System protection scheme SV Sampled value
UFLS Under frequency load shedding VT Voltage transformer
WAMS Wide-area measurement system WP4 Work package 4
Chapter 1
Introduction
As the global effort for the climate change increases, so does the demand for cleaner en- ergy production. Penetration of renewable energy sources introduces more power electron- ics equipment in the power system. Distributed power generation from these renewables causes a change from the traditional power system. These changes contributes to pos- sibilities and challenges which must be overcome to protect the integrity, reliability and utilization of the power system.
The traditional power system has large centralized power generators, and demands at other locations, usually in industrial, commercial and residential areas. Hence, the power flows from the generators to the consumers. The generators and loads are connected through the transmission system and distribution systems. As voltage levels vary from generation to the end customer, transformation is needed. This is done at substations. Here, mul- tiple lines are connected through busbars, acting as a connection point between different lines. The substations contains transformers for transforming voltages between the levels of the interconnected lines. Additionally, switches, monitoring and protection & control equipment are installed at these substations. Instrument transformers are used to measure voltages, currents and power in substations.
Different strategies are used to protect the power system from faults. Relays can be used to protect lines, busbars, transformers and generators. These relays are area protection devices which protects a certain element, for example a transformer, line or busbar, from damages if a fault happens. Failure to do so might also cause serious danger for the surroundings [1]. In substations, relays are installed in selectivity setups which is designed
such that protective action should affect the system as little as possible. Less selective protection acts as a backup service if the primary protection relay did not take appropriate action to clear the fault. Hence, these have intended time delay causing them to operate slower than the primary protection [2]. The relays utilize current and voltage transducer to pick up changes in the power flow. Traditional differential protection schemes relies on two conventional CTs and triggers if they measure different current levels.
1.1 Trends in the Power System
Industry 4.0 has set the path for digitization of the power system [3]. Such power systems are denoted smart grids. The change from traditional power systems to smart grids have different aspects. Benefiting from digitization, the smart grids utilizes Internet of Things (IoT) equipment so that every component in the power system can communicate. This enables data handling from different sources, that provides a cohesive and integrated view of the system being measured. Similarly, solutions are supported to increase situational awareness, and selection of optimal control actions to manage bottlenecks or disturbances.
Furthermore, as power electronics equipment is getting cheaper and manageable, such systems are installed in the power system to increase utilization, manage advanced electric machines and for dynamic control of scenarios. Flexible alternating current transmission system (FACTS) equipment are examples on power electronics equipment used to con- trol the power system. International connections, and offshore power generation is also growing, which introduces the need for High voltage direct current (HVDC) transmission systems. These are able to transfer power over long distances with lower losses than tradi- tional AC systems. These transmission systems requires a lot of power electronics equip- ment. Power electronics equipment reduces the voltage quality and could lead to decreased stability limits [4]. Lastly, renewable energy production like wind and Photovoltaic (PV) generation is altering how the transmission of power is flowing. Unlike the traditional power system, smart grids have distributed power generation which means power gener- ation can happen in the industrial, commercial and residential areas as well. The power system structure transitions from the traditional power system to smart grids is shown in figure 1.1. From these changes arises endless possibilities for improving the power sys- tem. This comes at a cost: The distributed generation, power electronics equipment, and introduction of new power generation creates challenges which has to be overcome.
As investments in renewable power generation grows, the net generation in the power system often increases. Especially in countries like Norway, renewable power production is installed to supply other countries instead of decommissioning carbon based production units. If new transmission lines are not installed, the stress on the power system increases.
More renewable production and HVDC links are being installed in the Norwegian grid [5, p.12]. These have low inertia, which means a change in load will cause a large change in frequency. Synchronous machine connected hydro plants have high inertia and gover- nor control which causes high damping of disturbances, giving stable protection against tripping of lines. Such systems are easier to control. Wind farms with low inertia, has low damping, and such disturbances will hence tend so oscillate, and give large frequency changes [5, pp.35-41]. In future power systems close to the shore, with offshore wind farms and international grid connections, the low total inertia can cause frequency chal- lenges. In interconnected power systems the effects of disturbances and faults can have
Figure 1.1:The traditional grid compared to IEEE’s vision of the smart grid structure [6].
impacts on the rest of the network far from the location where the disturbance or fault happened. For example, in 2003 trees short circuited a345kV line in Ohio, and as a re- sult multiple lines and generators disconnected in the area in order to protect equipment.
This caused a cascading effect, and ended in the blackout in the power system affecting 50million people in Northeast USA and Ontario, Canada [7]. A local fault hence affected a great part of the power system. To prevent this, trimming of the trees could be done.
However, even if safety precautions can be done, the risk of faults is always present. Some other examples on blackouts are Ukraine 2015, Venezuela 2019 [8], and India 2012 [9]. In such cases, the area protecting devices are not sufficient for protecting the power system.
SPSs were designed to protect bigger areas of the power system, utilizing special control schemes for protecting parts of the system from wide area faults.
The protection of the power system relies on transducers. Without these, the state of the system is unknown, and actions for controlling or protection the system is impossible.
Measurement systems operating quickly are necessary for determining appropriate action in protecting the system upon a fault. Industry 4.0 and the digization of the substations allows for new digial sensoring technology, and protection schemes. Fast communication solutions interconnect equipment and substations, but the performance of these communi- cations are dependent on latency, jitter, redundancy and cyber security [2].
1.2 Scope of Thesis
Digital substations offer benefits for the design of SPS. In the traditional substations, the complexity of cabling in connections made advanced protection systems hard to imple- ment, but the new digital substations offers endless possibilities for configurations. Inter- net communication means no new cables must be installed for a new protection algorithm to be implemented.
The reliability of these schemes are dependant on transducers, which motivates the re- search on the RC for use in differential protection schemes in digital substations. Tra- ditional substations utilize CTs, but the use of these is not without challenge. The iron cores means they are difficult to install in cramped locations, and they become costly when designed to operate without saturation during faults [10]. Hence, saturation of CTs is possible in practical installations. Saturation of CTs can cause maloperation of protec- tion relays [11, 12, 13]. In worst case scenarios CT maloperation can cause cascading wide effect faults. The research on RCs have therefore gained interest in the last years [14, 15, 16, 17, 18, 19]. The preeminent benefit is that the lack of an iron core means it does not saturate. The air core make it a cheaper option than the conventional CT. There are however, some challenges that most be overcome before the RC can be the primary option for protection applications transducers. Compared to the CT the RC is more prone to influencing quantities like conductor position, conductor angle, temperature and fre- quency deviations [20, 21, 22]. This influences affects the measurement accuracy, and this can result in malfunctioning of relays in protection applications. The RC is discussed more in-depth in chapter 3.
The background for this thesis is to advance the research on the RC for protection ap- plications. Several proposed methods of RC modeling exists. The lumped model, the distributed model and numerical models have primarily been utilized [23, 24, 25]. FEM models allow for more complex modeling compared to these methods. RC FEM modeling have been conducted in several publications [26, 27, 28]. The design of a FEM model is a powerful and realistic method for simulations which can be used to lay the groundwork for further work on the RC. Influencing quantities can be difficult to avoid, but if these quanti-
ties can be sensed using external sensors, then the measurements can be compensated and the measurement accuracy of the RC is assured in all operation situations. This in turn can ensure safe operation of the power grid during faults, and defend against wide area faults in the future.
The goal of the work is to test the accuracy of a RC in an appropriate test rig and build a FEM model to simulate the RC. This model is to work as a basis for the future devel- opment of compensating techniques. Compensating techniques utilizes sensors to obtain parameter changes in the RC configuration, and applies digital compensation of the signal to increase the accuracy. For example by using hall effect sensors to determine the conduc- tor position. The model can be used to investigate how different parameters influence the signal, and to determine valid compensating techniques. The possibilities for such a FEM model are endless, at the cost of computational power. Because of the geometry of the RC 3D models must be used, which quickly increases the complexity of these simulations compared to the 2D counterparts. Such models are however valuable as it offers benefits for future research into the topic. The research on protection applications can benefit from this work because a compensated RC might be able to work interoperable with a conven- tional CT in a differential scheme. This benefits the digital substations as upgrades can be done gradually.
The objectives of this thesis are the following:
• Develop a laboratory setup for testing the accuracy of a RC
• Test the influence of conductor positions on the RC in the laboratory environment
• Create an 3D FEM model of a RC
• Simulate conductor position changes and evaluate the effects compared to the labo- ratory tests
• Contribute to developments in RC modeling and accuracy compensation techniques using FEM model
• Contribute to protection applications interoperability research
• Contribute to the use of the RC in reliable SPS operation
Chapter 2
Power System Protection
As digital solutions are incorporated into the power system, new protection possibilities appear. The role of protection in the power system is to secure the reliability of power transmission. Reliable electricity supply is becoming increasingly important for the soci- ety, and pressures on the utilization of the power system is pushed as economic powers and renewable power generation increases [29]. System protection schemes, and digital substations offers possibilities to deal with these challenges.
The traditional substations lacks smart systems for communication in the substation. In the control building data from the substation can be sent to the power system operator. Com- munication within the substation consists of hardwired analog signals from CTs, Voltage transformers (VTs) and Potential transformers (PTs) to relays. The relay output are then transmitted analogous to Circuit breakers (CBs) and other relays [30]. For this communi- cation to work, a lot of hardwired cabling is needed, because a one wire must be installed between each communicating device. Additionally a wire must be run to the control house for monitoring. Configurations, repair and installment of equipment in such substations require a lot of effort. These analogous wires requires specific sizes and length for the accuracy of measurements, and additionally these can cause safety issues for personnel or equipment [31].
The introduction of digital substations enforces communication inside the substation on a peer to peer basis. This is assured by the process bus. The process bus is an Ethernet based communication channel between process level switching devices and bay level intelligent electric devices for protection and control applications [32]. Whereas in traditional sub-
Figure 2.1:An overview of an IEC 61850 based substation [37].
stations, signals were sent as analog electrical signals, the process bus transmits digital signals. Smart technology for sending and receiving data is installed on the process bus which allows uncongested data communication on the process bus. An apparent advan- tage of this is that the need for excessive cabling is reduced as all signals can be transferred through one or a few process busses. Additionally, control and measurements is electri- cally disconnected from the HV side of the substation, which decreases risk of injury [33].
Digital substations require time synchronization for all digital signals. This is usually done using a Global positioning system (GPS) satellite clocks. The clock signals are added to signals and measurements. Global synchronization is an requirement in Wide-area mea- surement system (WAMS) [34]. In order to utilize the benefits of the digital substation, the sophisticated communication of the process bus is required. In turn, this communication needs a robust communication standard. The solution is IEC 61850.
The IEC 61850 standard is made especially for substation automation. The project was started in during the 90s and was originally a cooperation between the US and EU (UCA and IEC) [35]. As corporations was growing more international a demand for standard- ized communication was growing. Units from numerous different providers meant it was troublesome to add functionality to the power system. The main goals of the standard is to give interoperability, free configuration and long term stability [36]. IEC 61850 sets rules for communication between intelligent electric devices in digital substations. IEC 61850
based digital substations employ the process bus for communication between devices. The process bus needs clear rules for transmission in order to work. Otherwise, packets would interfere and congest, and intelligent electric devices of different vendors will not be able to communicate. The benefits of IEC 61850 are [38]:
• Reduce dependence on multiple protocols
• Reduce construction cost by eliminating most copper wiring
• Automate substations
• Real time distributed computing
• Advanced management capability
• High speed peer to peer communications
• Improved security/integrity
• Flexible programmable protection schemes and ease of maintainability
A schematic of an example IEC 61850 based digital substation is shown in figure 2.1.
Switches between Infinite element domains (IEDs) on the process level facilitate support for peer-to-peer communication. Centralized control and monitoring can be done both at the substation, and in a control center at a different location. Configuration and installation of new equipment is easy, as all IEDs can be reconfigurated from a central unit through the switches. Peer-to-peer communication between units means transmission of data has greater performance, compared to the traditional setup where signals had to go through a central unit. IEC 61850 also includes support for multitechnology communication, which means relays are able to communicate directly with CTs, VTs and switchgear.
Generic object oriented substation events (GOOSE) is a functionality introduced with IEC 61850. GOOSE communication can happen on a peer-to-peer basis. This means the IED can communicate automatically between each other which significantly decreases the transfer time compared to a system where messages have to go through a central unit. This allows for quick circuit breaker action upon a relay trip in a substation. The GOOSE ser- vice is based on Ethernet multicast transmission which provides fast, efficient and reliable communication between the nodes [39]. Multicast addresses are addresses which multiple devices respond to. Hence an intelligent electric device only has to transmit one message to communicate with all subscribing intelligent electric devices. This design eliminates
the need for TCP/IP protocols, which reduces the complexity and time of transmission.
Additionally, the use of a GOOSE based process bus means a number of virtual ports can be utilized in an intelligent electric device which has limited input and output ports [40].
As the name implies GOOSE are object oriented messages. The benefit of this approach is
Figure 2.2: Normal condition (T0) and heartbeat messages (T1, T2, T3) for GOOSE packet trans- mission [32].
that all kinds of data can be transferred using the same protocol whether it is fault data, trip signals or breaker status.The functionality for different data types is what makes GOOSE object oriented, and offers great benefit for event messaging. For this to work, all IEDs in a substation must be configured to handle the type of data being transmitted. Depending on the setup in the substation, different types of data can be transmitted, sending only what is necessary. This reduces the size of GOOSE packages. GOOSE messages are sent out rapid intervals when an event happens in a substation. The transmission will be resent at a given frequency, which will be lowered until the maximum heartbeat frequency is reached.
After this, events are transmitted at a fixed, slower frequency. This mechanism is called heartbeat messaging, and enables fast transmission of events, and slower constant updates during normal conditions. figure 2.2 shows this method of transmitting event data.
IEC 61850-9-2 set the rules for Sampled value (SV) in digital substations. SV are used in for sending and receiving of current and voltage measurements on the process bus. In order to reduce the overhead of communication on the process bus, SV is directly mapped to the link layer under Ethernet type88 ba[39]. SV packets are created in Merging units (MUs), which digitizes current and voltage measurements, and time stamps the packets with a time stamp from the digital substation synchronization unit. The Ethernet protocol supplies destination and source Media access control (MAC)-addresses for where the SV is sent and received. Like the GOOSE protocol, SV also uses multicast MAC-addresses which means packets of both protocols can be transmitted on the process bus.
2.1 System Protection Schemes
SPS are defense plans against cascading effects in the power system. If such a fault hap- pens, the SPS will act and control the cascading effect - avoiding a potential blackout.
[41] argues that as complexity of network operation increases, because of growth in load, changes in market conditions and increased import and export. The network becomes more stressed, and hence SPS are growing in popularity. In power systems with low iner- tia it is harder to maintain frequency stability, which increases the risk of blackouts. The commissioning of new renewable power production decreases the overall inertia of the system [42]. This further emphasizes the importance of SPSs.
Generally SPS can be categorized as controllers having the following properties [43]:
• SPS can be armed or disarmed depending on system conditions
• SPS are “normally dormant” systems; initiating events usually occur less than once a year
• SPS usually employ discrete, feed forward control laws
• The control action taken is predetermined in most cases
• Typically some form of communication is involved in the control action
Even though the SPS systems are not often used, the blackouts exampled in section 1.1 shows the importance of these systems. These schemes can exist in different forms, with diverse goals. [44] presents these types of SPS:
• Generation rejection
• Load rejection
• System separation
• Turbine valve control
• Load and generator rejection
• Out-of-step relaying
• Discrete excitation controls
• Dynamic breaking
• Generator runback
• VAR compensation
• Combination of schemes
In the Hydro-Quebec area automatic measures are conducted upon predetermined events.
Switching of735kV shunt reactors to stabilize voltage levels done under extreme contin- gencies, generation rejection and load shedding done upon losing multiple735kV lines.
Underfrequency load shedding is performed to balance load and generation, and under- voltage load shedding is being studied [45]. Such SPS are usually triggered by a change in frequency or voltage. Depending on how much it changed, generation or load is tripped.
This is done to balance the system, and hence assure voltage and frequency stability. SPS can be simple in nature, operating in a local area, or they can be wide area based. A survey on the SPS was conducted in [45]. Here it was found that many of the existing SPS were based on local data. The local SPS can be simple and effective, but wide area protection offers a more reliable protection as measures for protecting the network is coordinated. In the example given on the 2003 blackout, the cascading effect in the network started the blackout. A local SPS could perhaps reduce the cascading effect when it reached the area.
However, if a wide area coordination SPS was implemented, the cascading effect would never start. Such schemes requires sophisticated communication between areas and inside substations. Traditionally such communication does not exist in substations. A system protection terminal is shown in figure 2.3. This presents a general overview on functional-
SPS Input Interface Local or Remote Signals
and Measurements
Local or Remote Control Signals
Power System Variables Database
GPS Time Sync.
SPS Output Interace
High Speed Communication
Interface
Supervision, Service, Maintenance and Update
Interface
Parameter Setting Database
Other SPS Devices Operator Interface
Power System Actuators Power System Transducers
and Measurement Devices
SPS Decision Making Logic
Low Speed Communication Interface Power System
Substation Control System
SPS Device
Figure 2.3:A system protection terminal [45].
ities of the SPS and interfaces to other SPS, the operator, measurements and actuators. The figure indicates that two concepts are of importance for the operation of SPS. Firstly, high speed communication between devices are needed to assure that the schemes can operate quickly enough when a fault happens. Secondly, power system transducers and measure- ment devices must be used to perceive such events. In a system of different vendors, and hardwired centralized communication, such schemes are not effortlessly implemented, as a lot of new cabling most be done.
In systems with multiple SPSs [43] found that inadvertent interactions between SPS could be catastrophic. From a process view the schemes seemed satisfactory, but when looking from a system point of view, the negative interactions between SPSs caused challenges.
In North America, it was found that in 70% of major blackouts relays contributed to the initiation or evolution of the blackouts [46]. [47] states that zone 3 distance protection re- lays misoperation has a major contribution to enforcing cascading effects. Zone 3 distance relays are backup protection devices which protects larger sections of transmission lines.
Hence, the need for smart relays, and system wide protection schemes are of importance in order to protect the power system, and to hinder the inadvertent interactions of SPSs.
Figure 2.4:A novel protection architecture in smart grids [48].
Some novel SPS takes advantage of WAMSs and uses this to trigger minimal intruding actions which stabilizes a wider area of the power system. Such schemes require inter- substation operation, and as such, relies on smart communication technology which is reliable, fast, and secure. With existing communication solutions for digital substations, the cost of interconnecting the communication over wider areas is insignificant [35]. Tra- ditional measurement systems uses local clocks for time stamping measurements, which
causes a challenge in calculating the exact system states [49]. To overcome this, novel SPS utilizes GPS synchronized time stamps. A novel protection architecture is shown in figure 2.4. In such a system, the protection of a substation is differentiated into areas.
Unit protection happens at a local area, with local measurements. The substation protec- tion measures faults inside the substation, and operates as a redundant primary protection when unit protection equipment is out of service. Further, the substation protection should work as an optimized backup protection which operates when the unit protection malop- erates. The wide area protection lastly uses WAMS, and cooperates with other substations to make decisions based on potential wide area disturbances [48]. This system should accordingly be able to protect against faults during normal conditions and during malop- eration, and stop cascading effects in the power system.
The use of novel technologies and approaches in SPS have been implemented and pro- posed. [50] uses a WAMS based Under frequency load shedding (UFLS) algorithm with promising results compared to conventional schemes. Artificial neural network (ANN) based UFLS schemes have also been proposed. [51, 52] used such algorithms for a SPS in a isolated system. These schemes are fast, robust and gives optimal load shedding values.
A drawback of ANN-based UFLS schemes is that they have to be trained on known data, and will hence have trouble with unknown cases [53].
As the power system increases in complexity, the need for SPS increases. In order to protect against wide area disturbances, smarter schemes are required. Computational in- telligence techniques can reduce the risk of blackouts [54]. WAMS have the potential for advancing SPS and its operation performance [44]. Sophisticated communication tech- nologies, IEDs and inter-substation cooperation is however required in order to take ad- vantage of these solutions.
2.2 Differential Protection
For protecting components in a power system, or switching out areas from faults, relays are used. These are protection equipment which uses electromechanical, static, digital or numerical technologies which triggers protective measures when a faulty event occurs [55]. The relays are dependant on transducer measurements. These measurements are read in the relay, and based on the technology used, an analog or digital trip signal is outputted if the measurements indicate that a fault has happened. Earlier, electromechanical relays were often used. These uses coils or temperature expanding metals to create mechanical forces with operates a relay contact. Although the electromechanical relays still exist today, more modern technologies are rapidly introduced, especially in high voltage power
Figure 2.5:Differential relay connection diagram [57].
systems [56] Digital relays depends on digitized measurements. These are then compared in a logic algorithm which based on the measurements outputs a boolean value which a CB acts on. An inherant benefit of digital relays is that different types of relay schemes can be configured. Examples of relay schemes are differential relays, differential relays and overcurrent relays. An electromechanical relay on the other hand, is designed to feature just one such scheme.
Differential protection is used to protect one or more components in between a protected zone. Current measurements from both sides,I1 andI2 are compared. Hence this relay demands that CTs are chosen appropriately. A differential protection scheme is shown in figure 2.5. In this figure the topmost inductor is the protected unit. Two CTs is situated on both sides of the protected unit.
Equation (2.1) is an expression of the differential current, denoted asI0 in figure 2.5 in relation to the transformed currents.
id=i1s−i2s (2.1)
ir=i1s+i2s
2 (2.2)
id≥Kir (2.3)
In equation (2.1)idis the differential current,i1sis the transformed current from CT1 in figure 2.5. i2sis the transformed current from CT2. A restraining current is used to make a comparison for the differential current. It is the average of the transformed currents. The restraining current is given by equation (2.2).
The percentage ratio,K, is typically set between10-40% [56]. Kcan hence be set ac- cording to the level of sensitivity wanted. It should cover measurement errors, ratio mis-
matches, margins and more. Additionally tap changing transformers must be accounted for when using such relays. Level detection is implemented when designing relays. These are configured by the pickup setting which is the minimal current where the relay will trip.
Figure 2.6 shows this effect. In order to trip this relay, a minimum current,1.0pu in the figure, and a minimal time is necessary to operate the relay. Interoperability of differen-
Figure 2.6:Characteristics of a level detector indicating pickup settings in relays. A minimal current and time is necessary to initiate a trip [56].
tial protection schemes poses challenges as equation (2.1) should net close to zero when in normal conditions. If the measurement have phase shift or magnitude differences, the differential current can be quite high even at normal conditions. The pickup setting should be configured to encompass this. Small disturbances, and external faults should not trip the relay, and must hence be considered as well. Conventional CT saturation could cause undesirable tripping of relays.
Figure 2.7:Overlapping protection zones denoted with different colors in a power system [55].
In SPS lots of relays are necessary to operate protective action over a wider area. In power
systems, relays are located such that the necessary units are protected. Additionally, to protect against maloperation of relays, backup relays are placed with a delay of operation such that these can protect the units of the primary relay failed. Protection areas are hence places such that protective schemes overlap each other. An example of this is shown in figure 2.7.
In conventional differential unit protection schemes, and as shown in figure 2.5, two con- ventional CTs are used to measure currents used in the relay operation. However, as Nonconventional instrument transformers (NCITs) are growing in popularity solutions for interoperable cooperation between CT and NCIT is of interest. Since differential protec- tion relies on equal measurements on both sides of the unit, such configurations could cause challenges as measurement accuracy, saturation, bandwidth, and phase shift could cause a non-zero differential current during external faults, steady state or disturbances in the grid.
2.3 Prodig Perspective on Digital Substations
The ProDig project is a research project. It is a cooperating between the Electrical Power Engineering and Information Security and Communication Technology departments at NTNU, UiO department of Technology Systems, SINTEF Energy Research and Michi- gan Technological University. Information on the project is found on the official website [58]. The project focuses on power system protection and control in digital substations.
Four PhD students and one PostDoc work on different aspects of digital substations and are organized into seven work packages. This thesis correlates with Work package 4 (WP4) which focuses on the use of NCIT and sensors in protection applications. The use of NCITs in differential protection interoperability configurations with conventional CTs is a research topic of interest.
The ProDig utilizes real time simulators, relay test devices, IEDs, process buses and satel- lite clocks to emulate a digital substation centered at the NTNU relay laboratory. This configuration, seen in figure 2.8, allows for practical tests of communication techniques, interoperability setups and use NCITs sensors. Parts of the goal is to investigate the in- teroperability of different sensors in a differential protection scheme utilizing IEC 61850 based process buses with GOOSE and SV communication. In order to use the RC in an interoperability configuration with CTs, initial research into the topic of RC is required.
One of the focus areas of WP4 is to utilize sensors to determine compensating sig- nals to correct a measurement if the transducer deviates from the standard condition. One
Figure 2.8:Schematic of technology used and intended interconnection of communication withing ProDig project laboratory resources [58].
example is to use hall effect sensors to determine the conductor position, and based on this apply compensation through the RC integrator. The accuracy testing laboratory setup designed for this work can benefit the project through the thesis and future tests performed by ProDig associates. This thesis contributes to the research project by creating data to work with and a model which can be developed with these compensating techniques. The project can use these results to do further testing in a mixed technology setup with a RC and a CT in a differential protection scheme using the ProDig laboratory.
Chapter 3
Rogowski Coil Modeling & Testing
The Rogowski coil (RC) is a NCIT, and was proposed in 1912 by W. Rogowski and W.
Tannhaus [59]. At the time the limited amplitude of the output meant it had confined ap- plications [20]. However, in recent years the RC has received new interest as the methods of creating the coil, and analyzing the output through Digital signal processing (DSP) has improved. Hence, making it a viable option in IEC 61850 based digital substations. The main benefits of the RC are the following [23]:
• Enduring large overloads without damage
• Measuring current in an extensive range, without saturation
• Easy to use, due to flexibility and light weight
• Low cost
• Non-intrusive nature (drawing no power from the main circuit)
• Wide bandwidth, in a range of0.1Hz to1GHz
• Excellent transient response
• Safety (isolated from the main circuit, electrically)
A RC is shown in figure 3.1a. The conductor in the figure is the conductor for which the current is measured. Here,vout(t)is the output signal of the coil. The lack of an iron core compared to conventional CTs gives the RC linear properties without saturation. Figure 3.1b shows a proposed solutions where the high current bandwidth and the non-saturation nature of the RC means one RC can replace multiple CTs in protection and metering applications as a current sensor or multi-purpose sensor.
(a)A RC structure. Here shown with an internal return wire loop [60].
(b)Proposed solution where one RC can replace multi- ple CTs in metering and protection applications [61].
In conventional CTs the open circuit voltage can be up to 10 kV. As the RC lacks an iron core, there is weak coupling between the primary and secondary side so a maximum voltage up to about10V will be applied to the secondary side even during faults. This makes the RC safe to operate, and removes the instrument security factor requirement [61].
3.1 Modeling & Output Integration
For a toroidal with an enclosed loop amp`eres law is given as I
C
B·dl=µ0 Z
S
J·dS. (3.1)
Bis the flux density,dlis a line element along the curvature of the coil denoted as the dotted line in figure 3.1a,Jis the current density in the conductor anddSis an area element, whereS is the area inside the coil. When a conductor is entrapped in the coil, the right side of equation (3.1) equals the total current flowing through the torus center. This results in
i(t) = 1 µ0
I
C
B·dl (3.2)
Further, asBandlis parallel in all locations along the torus, andB(t)varies with time only, the following relation can be found
B(t) = µ0i(t)
2πr (3.3)
whereris the radius to the center of the torus. In figure 3.1a, this is
r=a+b−a 2 Faradays law can be expressed as
vout(t) =−dφ
dt (3.4)
The sign of equation (3.4) can be ignored. Assume that the torus hasN windings along the torus axis. The change in total flux,φis then given by the rate of change of the total flux density through all the windings in the torus as expressed in equation (3.5). Equation (3.3) is used to relate flux to the current.
dφ dt =
Z B(t)˙ ·dS=AwB(t) =˙ µ0i(t)˙
2πr (3.5)
heredAis a area element in the winding, andAwis the area of a winding which is assumed to be constant when a RC is created with solid materials. Now, by combining equation (3.4) and equation (3.5), the current in the conductor can be expressed as
i(t) = 2πr µ0Aw
Z
vout(t)dt (3.6)
The constant coefficiant can be expressed as the sensitivity or mutual inductance, M of the measurement, and hence equation (3.6) becomes
i(t) = 1 M
Z
vout(t)dt (3.7)
Equation (3.7) is the expression for the current in a conductor in relation to the output voltage, measured by a RC.
Figure 3.2:A rogowski coil with an R-C integrator [23].
The Rogowski coil has an output voltage which is proportional to the change of current, as seen in equation (3.7). Because of this, the output signal must be integrated in on order to obtain the measured current. Different ways of implementing this integration exists.
The simplest integrator is an R-C integrator, shown in Figure 3.2. This integrator is a series connected resistance, and a shunt capacitor. With other words, a low pass filter.
This integrator works well for the system frequency and up to over10kHz [62]. Active analog integrators based on Operational amplifier (OpAmp) have also been proposed [16], [23], [62]. These have the benefit of increasing the bandwidth of the integrator, but for protection applications, this is not a major consideration.
When the output voltage is integrated, the signal is shifted 90 degrees so that it is in phase with the measured current. However, delays in the integration method could add a phase shift to the signal. This causes problems when using these signals for protection applications as the compared signals will be at different amplitudes compared to their peaks at a given time. The error increases with increasing phase shift. However in [62]
a small to zero phase shift was introduced by such a integrator at protection application frequencies. This effect is shown in Figure 3.3.
Digital integrators are simply microcontrollers or sophisticated machines with an algo- rithm which used DSP techniques to get an accurate output. Major advantages of these integrators is that DC-offset and phase shift can be dealt within the integrator. Compensat- ing techniques can also easily be included. In [63] it was found that the digital integrators have strong rejection of digitizing noise and increasing the signal resolution. These inte- grators uses integration methods. For example Simpsons method, the trapezoidal method or forward Euler method [64], [65], [66] [67], [68]. Digital integrators are well suited for
Figure 3.3:Bode diagrams of the two different integrators with the same value of R and C.G1(s) is the Measuring system (MS) based on RC with an RC integrator. G2(s)is the Measuring system based on an RC with an active integrator [62].
digital substations, as the signals should be digitized at some point anyways for use in the process bus.
3.2 Rogowski Coil Dependence on Influence Quantities
The measurement accuracy of the RC is dependant on its position in relation to the conduc- tor and external factors. These external factors can be frequency, nearby electrical fields and temperature. In protection applications inaccurate measurements can trigger unwanted trip signals, or fail to trip during faults. Hence it is important to access how these factors affect the RC measuring accuracy, so that they can be mitigated or compensated for.
Several experiments have been conducted to affection of these factors. In [69] the po- sition of the conductor is investigated in order to see how the eccentric positioning of the RC affects the accuracy. The authors conclude that it does affect it, but that it poses less of a challenge in protection applications as the accuracy dipped0.01%at approximately 1.5mm offset from center. It was also found that the discontinuity of the coil has an ef-
fect on the accuracy. Denotedδ, the discontinuity is caused by the difference in the first and last turn as shown in figure 3.4. Tests performed by the same authors in [70] showed that a discontinuity of15% introduced a significant error up to0.5%, however, it is not expected thatδis this high in real applications. In [22] it was found that a window type RC is a lot more resistant towards eccentric positioning of the coil compared to split core type RCs. How the RC is angled compared to the conductor might also cause variations
Figure 3.4: The discontinuity angle,β, to describe the angle difference between the first and last turn in the RC[69].
in the measurements. Three types of RCs is tested in the configuration seen in figure 3.5.
Ratio errors of these tests are about1%for a small split core RC and less than0.5%for a window type RC compared to the centered reference position. It is highly unlikely that the angle of the conductor is this high in real applications, if some consideration is done when installing the RC. The tests shows that if a RC is installed in a tight location, it is better to install a properly fitting RC which does not need to be angled compared to the conductor.
Temperature variations are investigated in [71]. According to the authors, choosing a correct value load resistance of the RC is beneficial, as it acts as a voltage divider with the copper of the windings. As the temperature increases, thermal expansion occurs in the coil which in turn decreases the resistance. Furthermore as the temperature in the coil in- creases, more electrons and phonons collide, increasing the resistance. Hence the change in coil resistance and the thermal expansion of the coil can cancel themselves out if the load resistance is chosen appropriately. The temperature dependence is seen in figure 3.6.
The accuracy is changed only in Parts per million (ppm) in relation to the temperature which indicates that variations in temperature within reasonable ranges should not affect the accuracy in protection applications in a significant way.
RC Conductor
Figure 3.5:Position of angled conductor in laboratory tests [22].
Nearby conductors generate electric fields, which can disturb the RC measurements in locations with lots of equipment. Substations is an example of such a location. Ideally, the RC should only react to currents in the conductor passing through it, but this is not necessarily the case. Two design choices exists to protect against external electric fields.
One possibility is to wound to opposite directed coils together, such that external fields are cancelled by the opposing coils. Another, easier method is to return a single wire through the RC as shown in figure 3.1a. It is argued that the internal return wire loop does the same job, but with an easier implementation [15]. In tests conducted by [71] less than0.005%
error is recorded for distances higher than0.4m from the RC was introduced from external conductors for a split core type RC.
Tests have also been done on RC in order to evaluate how multiple influencing quanti- ties affect the accuracy of the measurements. [22] studies the combination of conductor position, temperature and frequency variations combined. In these tests three types of RCs were tested; one window type RC, and two split core type RCs. A window type RC can- not be opened, which means the conductor must be passed through it. The split core type can be opened, and can hence easily be placed around a conductor. This however, has a disadvantage because it increasesδand usually means it is less circular. [22] finds that frequency deviations does not affect the RC measurements, but that a combination with different positions and high and low temperatures has a great impact on the measurements.
For use in protection applications, the RC measurements needs to be dependable under all conditions. If an inaccuracy is included in a RC in a differential protection scheme, it may cause inadvertent triggering or failure to do so under faults. Both cases would result in a malfunctioning differential protection scheme. Especially is the failure to trigger a fault under extreme contingencies a critical point, as the spread of a contingency can cause major damage to infrastructure, the environment or life. Hence, a RC used in protection
Figure 3.6:Temperature dependence in RC with different load resistances [71].
applications must work under all scenarios. It is important to gain an understanding on how influencing factors affect the measurements, and to introduce methods to compensate for the influencing factors.
To quantify the measurement accuracy of the RC, a standard is defined. In IEC 61869- 9 Digital interface for instrument transformers. It defines requirements for digital com- munication of instrument transformers with digital output [72]. Within IEC 61869-9 a definition for the calculation of measurement accuracy is included. This is called com- posite error and is used for all instrument transformers including conventional CTs. The composite error can be calculated as:
c(s) = v u u u u u u t
N−1
P
n=0
(iX(s−n)−iR(s−n)2
N−1
P
n=0
(iR(s−n)2
·100% (3.8)
Herec(s)is the composite error at a time instants.iX(s)is the measurement value of the instrument transformer ats.iR(s)is the value of the reference instrument transformer ats.
Nis the nominal sample rate in samples per second divided by the fundamental frequency in hertz.
To investigate the overall composite error of the RC, equation (3.8) can be modified
slightly to excludes, and hence becomes:
c= v u u u u u u t
N−1
P
n=0
(iX(n)−iR(n)2
N−1
P
n=0
(iR(n)2
·100% (3.9)
Using this formula, it is possible to compare the overall error of a RC compared to a reference CT.
3.3 Laboratory Setup
The RC accuracy tests are conducted in an ad hoc laboratory test rig designed with help from Mr. Katoulaei and the service lab at Department of Electrical Engineering. The test rig is seen in figure 3.8a. Two aluminium hollow conductors are fastened to the test bench
Figure 3.7:The Fluke i2000 flex used in laboratory testing [73].
through adjustable stands. These are placed on horizontal plates where the fasteners can slide to adjust the position of the conductors. Rulers placed to the side of the horizontal plate allows for millimetre adjustments of the positions and arrow indicators on the con- ductor stand clearly shows its placement relative to the ruler. The horizontal adjustment is shown in figure A.2. A RC in test is fastened to a frame which is held up by the test
rig. The RC is seen in figure 3.8c. It is connected through a cylindrical arm attached to a cylindrical stand which allows for vertical adjustments as seen in figure A.6. The use of cylindrical components allows for adjustments in how the RC is angled compared to the conductors. The frame height can also be adjusted by sliding the arm relative to the stand.
Protractors and arrow indicators are used to assist in adjusting the angles as seen in figure A.7 and figure A.8. The RC used is a window type Fluke i2000 Flex as seen in figure 3.7.
(a)The test rig. (b)The test devices. Left: Omicron, right: oscilloscope
(c) The RC frame. Position indicators glued to the frame, star contraption to verify adjustments which aligns with position indicators.
(d)Omicron CRC 356.
Figure 3.8:Photos from the laboratory setup and the equipment used.
It has a current rating of200A or2000A and an integrator. It has a large window suitable for high current conductors. The flexible build means it can be installed easily in a range of locations. However, because of this it does not hold a rounded shape well. It is locked by a clip, the black part of RC in figure 3.7 and figure A.5. Fluke does not advertise the number of turns or the discontinuity,δ. The accuracy is rated at±1%[74].
A Pearson Electronics Inc. current monitor referred simply as CT, shown in figure A.9, is used as a reference to compare the RC measurements and compute its error. The CT has an error of+1%according to the producer [75]. Both transducers are connected to
Figure 3.9: The RC model in orange with the air domain around the coil and the conductor in the middle of the figure. The terminals of the RC is seen on the right.
a InfiniiVision 2000 X-Series oscilloscope by means of coaxial cables. The test current is generated by an Omicron CRC 356 relay test set seen in figure 3.8d. The Omicron is able to output currents at different output levels and frequency32A banana plug cables are connected via the Omicron to the test rig conductors, figure A.4, hence limiting the output current to32A. A computer with the Omicron Test Universe software is used to configure the output of the Omicron. The Omicron, computer and oscilloscope is seen in figure 3.8b. A schematic of the laboratory setup is shown in figure A.1 displaying the connections without the test rig shown.
3.4 Finite Element Method Modeling Setup
A RC FEM model is built in COMSOL Multiphysics. It consists of a cylindrical con- ductor, enclosed by the coil. The coil is built from wound copper cables with a radius of rwire= 1.5mm. The coil has a major radiusR= 97.5mm and a minor radiusa= 5.6 mm. The minor radius is the radius of each winding. The radius of the conductor is5mm.
For the coil, the return wire concept is utilized to protect against external fields.
Parametric curves are drawn in the model, and represented by three expressions in three dimensions
